Local Stability and Region of Attraction Analysis for Neural Network Feedback Systems under Positivity Constraints
Published in IEEE Conference on Decision and Control (CDC), 2025
IEEE Xplore · arXiv · Code · Libraries: nnsector, nnstab
In short
How large is the set of starting states from which a system with a neural network controller still settles at its equilibrium (its region of attraction)? For positive Lur’e systems, the paper gives two ways to estimate it: a Lyapunov method with a linear matrix inequality (LMI) that grows invariant sets, and tight local sector bounds for the network, computed layer by layer and used in a localized Aizerman test. In the examples, both give larger regions and scale better than methods based on integral quadratic constraints (IQC).
Topics: region of attraction, local stability, neural network verification, local sector bounds, linear relaxations, LMI, IQC.
Abstract
We study the local stability of nonlinear systems in the Lur’e form with static nonlinear feedback realized by feedforward neural networks (FFNNs). By leveraging positivity system constraints, we employ a localized variant of the Aizerman conjecture, which provides sufficient conditions for exponential stability of trajectories confined to a compact set. Using this foundation, we develop two distinct methods for estimating the Region of Attraction (ROA): (i) a less conservative Lyapunov-based approach that constructs invariant sublevel sets of a quadratic function satisfying a linear matrix inequality (LMI), and (ii) a novel technique for computing tight local sector bounds for FFNNs via layer-wise propagation of linear relaxations. These bounds are integrated into the localized Aizerman framework to certify local exponential stability. Numerical results demonstrate substantial improvements over existing integral quadratic constraint-based approaches in both ROA size and scalability.
BibTeX
@inproceedings{montazerihedesh2025local,
author = {Montazeri Hedesh, Hamidreza and Wafi, Moh Kamalul and Siami, Milad},
title = {Local Stability and Region of Attraction Analysis for Neural Network Feedback Systems under Positivity Constraints},
booktitle = {2025 IEEE 64th Conference on Decision and Control (CDC)},
address = {Rio de Janeiro, Brazil},
pages = {1142--1149},
year = {2025},
doi = {10.1109/CDC57313.2025.11312171}
}
Recommended citation: H. Montazeri Hedesh, M. K. Wafi, and M. Siami. (2025). "Local Stability and Region of Attraction Analysis for Neural Network Feedback Systems under Positivity Constraints," 2025 IEEE 64th Conference on Decision and Control, Rio de Janeiro, Brazil, pp. 1142-1149.
Download Paper
